The nuances of speech recognition

In this interaction, Sunny Rao, managing director, India and South East Asia, Nuance Communications, tells Sudhakaran of CIOL about the speech recognition software, its applications and market potential. Excerpts:

How do you look at the speech recognition software market globally?

With more than 6,000 people and sales that are expected to reach US$1.2 billion in 2010, Nuance is the leader in speech technologies generally. Our market includes not just software, but also services and systems, in areas ranging from desktop software, to call centres, healthcare, mobile phones and automobiles. It’s a diverse and growing market. Speech is probably the preferred way for people everywhere to interact with technology.

Which are the areas where speech recognition software comes to the help of users?

Key application areas of speech recognition software include medical and legal reporting, field reporting (insurance claims, field sales, field service, social work, inspections), intelligence reporting and incident reporting in law enforcement, RSI (Repetitive Strain Injury) prevention/treatment, providing accessibility to technology for people with physical handicaps and assisting people in overcoming language-based problems, e.g., dyslexic symptoms, call centre agent wrap-up, education etc.

Most people type at a rate of 40 words per minute but can speak at a much faster rate, clearly demonstrating that dictation offers obvious productivity advantages for tasks like report creation. In fact, Dragon NaturallySpeaking enables users to create documents and e-mails three times faster than typing and delivers industry-leading accuracy rates of up to 99 per cent.

Can you describe the function of the software? Can it convert text into voice as well?

Dragon can read back text to you using Nuance’s Vocalizer technology. Its version 11 delivers incredible accuracy and speed that people have come to rely on. It includes many more features to help people speak their minds. Dictation is just the tip of the iceberg when it comes to driving higher workforce performance and productivity with speech recognition software. Dragon NaturallySpeaking also helps users of all kind, be it professionals, students, teachers, writers, bloggers, or even common people, to get their thoughts on paper as fast as they can articulate them in spoken words, including emails, documents, spread sheets and presentations.

In today’s information-driven workplace, managing e-mail takes up an increasing amount of time. With Dragon NaturallySpeaking, users can create, navigate, send, and respond to e-mail — all by voice — using popular programs like Microsoft Outlook or Lotus Notes.

Also, Dragon NaturallySpeaking Professional enables users to navigate their computer desktop, controlling virtually any menu item or dialog box entirely by voice for convenient, hands-free operation. Users can edit and format their work, launch applications and open files, navigate forms by voice, and insert standard blocks of text to dramatically speed up routine tasks on the PC - even when they’re on the move.

Which are the languages that Nuance Dragon can recognize?

Dragon is available in American English, Australian English, Asian English, Indian English, UK English, Dutch, French, German, Italian and Spanish. The French, Italian, German and Spanish editions also support English. And the Dutch edition supports English, French and German. In India, we have already rolled out the version of the software that recognises all Indian English accents.

In a country like India, where there are many accents, what is the accuracy level of Nuance Dragon?

As I said earlier, Dragon software recognises all Indian English accents. Within seven minutes of training, the user can get up to 99 per cent accuracy.

Can it be custom made to 'understand' a particular accent or style of an individual?

Absolutely! You can teach Dragon to make it more accurate. The more you use Dragon and correct any errors it makes, the better is the accuracy. Regular usage of the software helps Dragon to recognize your voice and it becomes more accurate by improving the recognition response. Dragon Premium and Professional versions can also be customized with your own commands to insert commonly used text or, in the case of professional, do complex processing based on your commands.

What are the improvements that you have made in the Nuance Dragon Naturally Speaking Version 11?

As compared to the previous Dragon versions, Dragon 11 uses advanced Technology, which has made it smart and accurate in speech recognition. Despite the sophisticated nature of speech recognition, Dragon’s 'brain' works behind the scenes, so you can focus on creating and communicating by voice at speeds up to three times faster than typing — without the software getting in the way. It boasts of an improvement in accuracy of up to 15 per cent compared to Dragon 10, and it is also faster than previous editions when selecting application menu items by voice or executing voice commands.

The time-saving voice commands and shortcuts of Dragon collapses common multi-step tasks on the PC into direct voice commands. Morevoer, the updated toolbar allows users to discover and quickly access important but often-overlooked Dragon features.

Also the new version makes editing and correcting text easy. With Dragon’s Quick Voice Formatting commands, you can issue a simple voice command to make formatting text faster than ever. With Dragon 11, it’s easy to edit or correct a word or phrase if there are multiple matches by displaying a number next to each. In addition, corrections entered using the keyboard (in addition to those made by voice) are used to adapt the user’s profile and boost accuracy. Finally, the Correction Menu now suggests more alternative recognitions by default and enables users to quickly add phrases to the vocabulary, or to prevent an undesired word from being recognized.

Can it be used to transcribe recorded audio?

Yes, with Dragon, you can dictate on the go into a digital voice recorder, and then let Dragon do the work for a quick transcription of your recorded voice. Creating a new user profile for a digital recorder or PDA, or adding a digital recorder as a new audio source for an existing user profile, is now much faster with Dragon 11.

Dragon 11 is compatible with Microsoft Office 2010 applications and Full Text Control, Menu Tracking, and Natural Language Commands are supported for Microsoft Word 2010, Microsoft Outlook 2010, and Microsoft Excel 2010.

How is Nuance Dragon integrated with mobile phones?

Mobile phones with suitable software may be used as digital recorders to provide speech for the Dragon system to recognize. Separately, Nuance delivers Dragon Mobile Applications in other parts of the world, allowing folks to simply speak and have their words transcribed for use in text messages, emails, and social networking sites like Facebook and Twitter.

What is your foothold in India and who are your major clients here?

We are now strongly entrenched in the Indian market and have major clients in legal, healthcare, education and government space.